The Pakihi Bog Began As a Lake When the Surrounding Area Was Forested by Both Podocarps and Silver Beech

76 PROCEEDINGS OF THE NEW ZEALAND ECOLOGICAL SOCIETY, VOL 22, 1975 A LOWLAND VEGETATION SEQUENCE IN SOUTH WESTLAND: PAKIHI BOG TO MIXED BEECH-PO DO CARP FOREST PART 1: THE PRINCIPAL STRATA

A. F. MARK and P. M. F. SMITH

Botany Department. University of Otago. Dunedin, New Zealand

SUMMARY: Six stands of lowland vegetation representing transitions from pakihi bog to mixed beech-podocarp forest are described quantitatively from the coastal plain of the Arawata and Jackson Rivers in South Westland. The sequence depicts progressive changes in canopy height. biomass and floristic richness. and appears to involve, at least in part. a primary succession. Some boundaries have been sharpene:i by periodic fires but the long+ term trend remains unobscured. Pollen evidence indicates that the pakihi bog began as a lake when the surrounding area was forested by both podocarps and silver beech. The bog seems never to have supported a \'.'oo:l.y vegetation but in the absence of fire. forest would slowly encroach on it. With a highway traversing the sequence. its reservation for educational and scientific pur~ poses has been proposed.

INTRODUCTION flat or sloping) type of Paludification (blanket) Bog (Cranwell, 1952). They are characteristically under- A feature of lowland coastal gravel plains and lain either. by peat or by highly leached soils asso- terraces throughout Westland is the occurrence of ciated usually with a loess cap and a strong iron pan openings on poorly drained sites in otherwise ta1l formation which seriously jmpedes drainage. lowland forest (Fig. I). These forest openings or On pakihis not recently disturbed by fire there is pakihis have been classed as a soligenous (surface a well defined sequence of vegetation. The open central part of the bog merges into a woody margin via open scrub which gradually thickens and increases in height from low shrubland through woodland, culminating in tall lowland forest (Fig. 2). The concentric pattern of vegetation types representing this sequence is strongly suggestive of a natural plant succession, as indicated by Cockayne (1928, p. 180). Holloway (1954, p. 381) and others (Chavasse, 1962). Indeed, Holloway (p. 381-2), referring to the patchiness of lowland Westland vegetation, suggests that the pakihi succession is part of a more widespread vegetation response to cJimatic changes occurring some time about the twelfth century A.D. He concedes, however, that the more open pakihis may be stable or at best represent a very slow succession because of recalcitrant soil FIGURE 1. View southeast from the slopes of Mt. conditions. McLean showing the lowland forest and pakihi of In contrast, there is evidence in some pakihis at Dismal Swamp about 2 km northeast of the study least, that they may represent the reverse situation, area. Hindley Creek drains the valley and crosses that is, a forest retrogression associated with soil the pakihi on the far left. May 1965. degradation following a period of apparent steady MARK AND SMITH: SOUTH WESTLAND LOWLAND VEGETATION; PART I 77

The southernmost pakihis may be those on the extensive coastal plain between the Waiatoto and Arawata Rivers which, being south of the Paringa River, are associated with a mixed beech-podocarp forest. This paper describes the typical vegetation sequence in the vicinity of Dismal Swamp, one of the most extensive pakihis on this plain (Fig. 3). The study began as a University of Otago Science Students' project in 1965 (Smith, 1965) but additional information has been collected subsequently. This paper describes quantitatively the principal strata represented in the vegetation sequence. The epiphytic and ground vegetation, in which bryophytes predom- inate, is described separately as Part 2 (Scott and Rowley. 1975). FIGURE 2. Vegetation sequence along pakihi edge in the study area at Dismal Swamp. Open short manuka THE STUDY AREA scrub in the foreground grades into increasingly Physiography tall dense manuka woodland, followed by silver pine (dark rounded canopies) with young podocarp forest, The area studied is underlain by gravel, sand or especia'ly rimu, silhouetted on the skyline. May silt (Mutch and McKellar, 1964). It is situated about 1965. 4 km inland and ]2 m above sea level. Since the highway immediately north of the Arawata River bridge traverses the entire vegetation sequence, sites state characterised by tall lowland forest. This for sampling were selected adjacent to it (Fig. 3). process, through developing a highly infertile gley The highway not only facilitated access to this low- podzol with associated iron pan, results in soil con- lying and featureless country but also ensured ditions inhibitory to the lowland forest and allows minimal interference by deer. the establishment of silver pine (Dacrydiul11 A small isolated outcrop of undifferentiated colensoi)* (Stevens, 1968). Rigg (1962) even claims metamorphic rock which the highway crosses about that pakihis occupying glacial and interglacial I km north of the bridge (Mutch and McKeIJar, terraces in the vicinity of Westport, north Westland, 1964) has somewhat higher relief and was avoided. have developed largely from forest due either to soil deterioration under forest or to climatic change, Climale or both. Further, she suggests the possibility of Official Meteorological Service records for two several cycles of succession and retrogression since nearby coastal stations, Jackson Bay] I km to the most of the larger pakihis contain buried timbers. north-west and Haast, located 29 km northwards However, she admits that the successional phase along the coast, indicate an annual rainfall somewhat is now widespread and has been "for many in excess of 3,500 mm fairly evenly spread over centuries", although recent burning together with about 200 days, and a mean annual air temperature increased waterlogging due to reduced water con- of about 11°C with relatively slight diurnal and sumption following logging, have counteracted this seasonal fluctuations. Screen frosts occur on 3 to 9 tendency at least in the Westport area. This latter days between May and October while ground frosts claim, however, finds no support from McDonald's may occur on 53 to 70 days throughout the year. Fog (1955) comparison of soil moisture content and and snow are both uncommon (12 year means of 8.5 other physical properties between forest and adjacent and 0.5 days, respectively, for Jackson Bay). sites reverting to pakihi following logging operations. Despite the large number of rain days, sunshine duration (for Haast) is relatively high-I,816 hours * Nomenclature follows Allan (1961) for pteridophytes, annually or 43 % of the possible total. gymnosperms and dicotyledons, Moore and Edgar (1970) for monocotyledons except for the grasses which follow Cheeseman (1925) or where applicable, METHODS Zotov (1963). Otherwise authorities are given in the species list (Appendix, Table I). Voucher specimens of Six sites were chosen to represent the vegetation most species listed are deposited in the OT A sequence from bog near the centre of the pakihi to herbarium. fully developed lowland forest close to the Arawata 78 PROCEEDINGS OF THE NEW ZEALAND ECOLOGICAL SOCIETY, VOL. 22, 1975

LOCA rJON '..' " .j"" '

Stond location "'" Moture lowland.jor'est (hilly)

Malure lowland forest (flatland)

Young rimu forest /' Silver pine wOOdland

Monuko Shrublancl

'09

FIGURE 3. The coastal plain between the Wala/olo and Arawata Rivers, South Westland, showing distribution of the main vegetation types and the approximate location of six stands studied to describe the vegetation sequence. The flatland east of the highway is generally known a.v Dismal Swamp, that to the west as the Burmeister Morasse. The map was compiled from an enlargement of aerial photograph 3977/11 supplied by N.2. Aerial Mapping Ltd., Hastings, and with a.fsistance of an unpublished National Forest Survey forest type map supplied by the New Zealand Forest Service.

River (Fig. 3). For all except the herbaceous plant (iii) Shrubs: woody species 0.3 to 5 m tall (composi- cover on the pakihi the point.centred quarter method tion, density); of plotless sampling (Cottam and Curtis, 1956) was (iv) Subshrubs: woody species < 0.3 m tall (compo- used, as for previous studies involving a similar range sition, density); of vegetation (Mark, et at., 1964). (v) Herbs: herbaceous species> 15 em tall (com- The categories (and values recorded for each) were: position, density). , (i) Trees: woody species> 10 cm d.b.h. (composi- Fifty points were used in all except the mature tion, density, basal area);' forest stand where the number was doubled for the (ii) Small-trees: woody species < 10 cm d.b.h. but> small-tree, shrub and suhshrub layers to cope with 5 m tall (composition, density); their increased floristic diversity. Points were distri- MARK AND SMITH: SOUTH WESTLAND LOWLAND VEGETATION; PART 1 79

STAGE

QRACOPHYlLUM LONGIFOLJUM

LEPTOSPERMUM SCoPAR1UM

QACRYOJUM COLEN SO!

PHYLLoCLAOUS ALPINUS

CoPROSMA FoETJDlSSJMA

COPROSMA COLENSO!

NEOMYRTUS PEOUNCULATA

NEoPANAX COLENSO!

MYRSINE DIVARICATA

DACRYDIUM CUPRESSJNUM

PODDCARPUS HALLI!

NOTHOFAGUS SOL v CLIFF

DACRYCARPUS DACRYDJOIOES

PSEUOOPANAX CRASS!FOLJUS

WEINMANN!A RACEMOSA

NOTHOFAGUS MENZIES!!

ELAEOCARPUS HoOKERIANUS

DlCKSONIA SOUARROSA

POOOCARPUS FERRUGINEU5

PSEUDOWINTERA COLORATA CANOPY HEIGHT 1m) ) , 0.' " " " Subshrubs Novalu. Novalu. 3~.SOO 3QOO 7,100 TOTAL 85.000 Shrubs 6,500 53,~00 13,900 22.900 31.~00 6.000 DENSITY Small"trus - - 5..70 ..0'30 2.800 1.030 (sttms/ha,) { Trus - 1,130 1,OeO '" TOTAL BASAL AREA -T,us - - I, "~9 ~5S9 6.551 (dm'ilha1

FIGURE 4. Values for relative density of the important woody species (>5% in one or more stands) in four size claj'ses (see key) from six stand... of vegetation (StaRes ]-6) at Dismal Swamp, South Westland. Values entered for subshl'ubj. in StaRes] and 2 are re/ative.J::Qver values. Relative basal area of tree species is also included (open blocks) while canopy heights and stand totals for density and basal area are shown at the bottom. 80 PROCEEDJNGS OF THE NEW ZEi\LAND ECOLOGICAL SOCIETY, VOL. 22, ]975 buted at 20-pace intervals along compass lines of buried wood intercepted during the sampling was 3000M except for the stands (2A) where lines were noted and where possible collected for identification. selected so as to remain within the limits of the The area was revisited in 1972, about five years particular stand. For the herb layer on the pakihi, after an extensive summer fire had !o.wept through estimates of percentage cover were more practicable. the bog and woodland sites previously studied. Sey. Here 50 quadrats, each 0.5 m\ were distributed as eral other pakihis more distant from the highway for the quarters. were also visited at this time and brief notes made Values for total and relative basal area and density on their vegetation. were calculated as for previous studies (Mark et al., 1964; Wells and Mark, 1966). For practical reasons, any stem of bog dacrydia species emerging from the RESULTS ground was assumed to be a separate individual and I. The Vege!ation Sequence thus eligible for sampling even though it may have been adventitious in origin (Moar, 1955). QuantHative values for the more important species Subsequent to completion of our sampling pro- along the vegetation gradient are shown in Figures 4, gramme the quarter method which we used was 5 and 6, while a complete floristic list is given in found to give abnormally high values for tree basal the Appendix (Table I). Growth ring counts of area (Franklin, 1967), particularly in stands with a manuka (Leptosperl11ul11 scopariul11), DracophyllunT wide range of stem diameters (Mark and Esler, 1970). longifolilinT and silver pine stems from the four Our absolute values for total basal area may there- youngest stands are summarised in Table 1. fore not be reliable but the relative values should A brief description of each of the six stages studied be satisfactory. is given below: A soil corer was used to remove samples of peat from two depths (0-20 em and 180-200 em) for deter- Stage I: The Pakihi mining colour, organic matter and pH. The organic content was measured by igniting oven dried samples The bog surface, although essentially level, consists at 7000C for two hours and acidity with a Macbeth of a network of minor depressions alternating with Model 1051 pH meter using 1:1 by volume dilutions low peaty mounds up to 30 em high and 1 m across. with distilled water. The colour of moist soil was Ca.'orophus minor (wire brush) dominates the bog checked against Munsell's standard soil colours. surface, being rooted chiefly on the mounds, but its The bog surface was surveyed along a line from procumbent habit allows it to partly conceal the near its centre (site 1 in Fig. 3) to the margin with depressions. It therefore provides most (90%) of the a Watts Autoset level and staff, and a Hiller corer 58'){, plant cover in the canopy layer (> 15 em tall). was used to measure depth of peat at intervals along Gleichenia circinata (tangle fern) and the rush this line and also to obtain a bottom sample from Baumea teretifolia are both common but, being quite its deepest part for pollen analysis. Location of any erect in their habit. make only a small contribution

TABLE t. Basal diameters (em) and numbers of growth rings in stems ofLeptospermum scoparium (manuka), Dacrydium colensoi (silver pine) and Dracophyllum longifolium (inaka) from four stands on or near to the pakilzi, Dismal Swamp, South Westland. Means of 4 to 7 determinations and their standard errors only, are given. (Value without a S.£. is a single determination.)

Species Stand J Stand 2 Stand 3 Stand 4 Pakihi Young manuka Old manuka Silver pine No. No. No. No. Diam. S.E. rings s.E.1 Diam. S.E. rings S.E. Diam. S.E. rings S.E. Diam. S.E. rings S.E. --- Dracophyllllm /ongifolium t.15 0.11 15.2 1.5 ------Leptospermllm scoparmm 0.67 0.01 15.4 t.1 1.26 0.04 18.7 1.0 6.68 0.51 37.9 1.0 ------Dacrydium colensoi ------8.9 - 220 - 25.5 0.8 339.4 9.2 --- MARK AND SMITH: SOUTH WESTLAND LOWLAND VEGETATION; PART 1 81

STAGE

CALAROPHUS MINOR BAUMEA TERETIFOLIA BAUMEA TEN AX GLEICHENIA CIRCINATA GAHNIA RIGlDA BLECHNUM MINUS BLECHNUM CAPENSE UNCI N1A RUPESTRIS HYMENOPHYLLUM SCA BRUM ASTELIA FRAGRANS MICROLAENA AVENACEA BLECHNUM OlSCOLOR

TOTAL DENSITY (stems I ha.)

FIGURE 5. Values for relative density of the important species (>5% in one or more stands) in the herb layer for the same six stands described in Figure 4. Note, va.'ues for Stands J and 2 are percent cover.

to the plant cover (Fig. 5). several bryophytes-for more detail see Scott and Co-dominant and emergent shrubs of manuka Rowley (1975). and Dracophyllum longifolium are scattered through most of the bog. Their small size is indicated by their Stage 2: Young Manuka on Pakihi Fringe minor cover values (1.3% and 0.5% respectively) The most striking change from Stage 1 is an provided by about 6,500 shrub-sized plants per hect- increase in density of manuka shrubs (from ca 3,000 are (2,950 of manuku and 3,550 of Dracophyllum). to 53,400 per hectare) but their canopies remain Five plants of each species in this stand had sufficiently small and open to allow persistence of similar numbers of growth rings (15) with less a typical pakihi community (Figs. 4 and 5). Only a variation among the manuka (Table 1). Basal dia- few minor species have disappeared while others meter (minus bark) of the manuka stems was less have established (see Table I in Appendix), especially than half that of Dracophy/lum (Tahle I) but on all on the mounds which here may reach 50 cm high. manuka shrubs examined, the bark at and beneath Large tussocks of Gahnia rigida are conspicuous the soil surface was noticeably inflated. Thus, of the here and there and there are occasional seedlings of five stems sampled, mean thickness of bark at the silver pine and Phyllocladus a/pinus established on ground surface was 1.19 + O.19;mm which was the drier mounds. 29.2 '){, + 2.9% of the total stem diameter, com- Typical manuka shrubs in this stand contain.1, about pared with 0.29 + 0.04 mm and 8.6 % J: 1.5 % for 20 growth rings (Table I) but a few standing dead the same stems about 20 cm above ground level. charred sterns of silver pine among the low manuka Most depressions in the bog surface were partly scrub indicate a history of fire. water-filled at the time of study-it contributed 22% to the ground layer-but the peaty surface beneath Stage 3: Manuka Woodland carried a sparse plant cover. Only the mosses Dicranoloma billardieri and Campy/opus introflexl/s This zone, up to 20 m wide, is characterised by a are general1y important in these hollows, contributing dense canopy, 3-6 m high, of small trees (5,470 together about 66 % ground cover but several other stems per hectare), chiefly manuka (Figs. 2 and 4). species may be - locally important or play minor Several average~sized manuka stems from various roles, e.g. Lycopodium ramulosum, HerpoUron parts of the stand contained about 40 growth rings novae-zelandiae, D rosera spathulata, D. binata, while those of silver pine, which contribute to % of Nertera depressa, Gaimardia setacea, Liparophyllum the small-tree stems and occasionally reach tree size, gunnil, Thelymitra venosa, Euphrasia disperma and are much older (Table I). PROCEEDINGS OF THE NEW ZEALAND ECOLOGICAL SOCIETY, VOL. 22. 1975

S1 I. DACRVDIUM S1. 5 DACRVDIUM CQLENSQI COLENSQI (157) (36)

51. 5 DACRVDIUM ST. 5 WEINMANNIA ST. 5 NOTHOFAGUS CUPRESSINUM RACEMOSA MENZIESII (87) (33) (22)

5T. 6 DACRYDIUM S1. 6 WEINMANNIA 51. 6 NOTHOFAGU5 CUPRE551NUM RACEMOSA MENZIESII (1.5) (61) (1,1)

FIGURE 6. Percentage distribution of diameter size classes for trees (>10 em diam.) of four species in three stands (St. 4~6). Only species with more than 20 stems measured per stand are included-values in brackets after the species names refer to the number recorded. Note that the size clas.5 intervals are 2.5 cm for Stands 4 and 5 and 5 em for Stand 6.

Manuka also predominates among the shrub and Dracophyllum longifolium, however, is reduced to subshrub layers but small stems of silver pine are a minor role. Gleichenia circinata is the most ~on~ plentiful. Several typical forest species have also spicuous herb with less of Baumea spp.. while become established (Fig. 4). Apparently young, Calorophus barely persists. Blechnum, minus and B. vigorous stands of rimu (Dacrydium cupressinum) vrocerum enter the sequence at this stage but play and kamahi (Weinmannia racemosa) are common, only a minor role (Fig. 5). with fewer of broadleaf (Griselinia littoralis), southern rata (Metrosideros umbellata), Neomyrtus Stage 4: Silver Pine Woodland pedunculata, Coprosma foetidissima and C. colensoi. The manuka woodland grades fairly abruptly into MARK AND SMITH: SOUTH WESTLAND LOWLAND VEGETATION; PART I 83 a woodland about 12 m tall dominated by tree-size 0ccupied almost entirely by immature stems of rimu stems with a total density of about 1,130 trees per (Figs. 2 and 7), even though it contributes only hectare. Only silver pine is conspicuous in the can- 44'!i, to the 1,080 tree stems per hectare (Fig. 4). opy (78.5 ReI. Den.; 88.8 ReI. Basal Area-Fig. 4) Their trunks, however, comprise almost 60 % o( the with stems of the most frequent size class (Fig. 6) total basal area and moreover show a fairly uniform containing about 340 growth rings (Table I). Several spread of diameter up to 5 dm (Fig. 6). Stems of forest species reach tree size but all have smaIl, silver pine, the second most important tree species, apparently young stems «20 cm d.b.h.) and are of have been overtopped by rimu and it is more poorJy minor importance-they include rimu, mountain represented among the smaller size classes than in beech (NothofaRus solandri var. cliffortioides), the woodland stage (Fig. 6). Phyllocladus alpinu,!l, Hall's to tara (Podocarpus hallii) Forest tree species continue to increase in density, kamahi and silver beech (N. menziesii). Manuka is with predominantly small, vigorous, apparently reduced to a minor role among the trees but over- young stems (see values for kamahi and silver beech topped stems persist as the most numerous small- in Fig. 61. tree species (23.5 ReI. Den.), ahead of establishing Occasional moribund shrubs of manuka persist, sterns of mountain beech (20.5%), Phyllocladus but much more numerous are those of. typical forest 07;{,), kamahi (10.5%) and rimu (8'){,), and supressed stems of silver pine (9%).

FIGURE 7. The young rimu stand exposed by the hiRhway close to the area sampled. Dominance of tall thin stems of rima and a well-stocked understory are apparent. The earlier stages of the sequence are in the background. October 1972.

In both the shrub and subshrub layers the role of manuka has been superseded by Phyllocladus, Neomyrtus and silver pine, while several other typical forest species have become established (Fig. 4). Blechnum minus predominates among the herbs while C.'eichenia circinata, Baumea gunnii and Gahnia rigida are less conspicuous (Fig. 5). Distribution of diameter size classes among the silver pine trees in this stand (Fig. 6) indicates a deficiency of smaller stems, suggestive of a decline in regeneration potential. '

Stage 5: Young Rimu Stand FIGURE 8. 1nterior of the mi:i:~d for~st stand near , the area sainpled. Blechnum discolor is the promin- The canopy of this stand reaches 20 m and is ,ent fern in the foreground. 84 PROCEEDINGS OF THE NEW ZEALAND EcOLOGICAL SOCIETY, VOL. 22, 1975 species: Neomyrtus. Coprosma joetidissima, C. ably a factor here since several palatable species colenso;, Pseudopenax colensoi, Elaeocarpus hook- are present (Coprosma spp., Pseudopanax colensoi), erianus (pokaka). Among the herbs only Blechnum while pepper tree (Pseudowintera c%rata) remains minus is important (Fig. 5). insignificant (5.7 ReI. Den. among the shrubs). Density of herbs is reduced to about half that Stage 6: Mature Lowland Forest recorded in any other stand, but among them Blechnum discolor has increased to be as abundant Several lowland podocarp species-rimn, kahikatea as B. minus, while B. capense, Microlaena and (Podocarpus dacrydioides), miro {Podocarpus fer- Astelia nerl!osa are much less numerous (Fig. 5). rugineus)--reach ca. 30 m and emerge above rather separate strata of silver beech and kamahi. The 2. Soils important tree species all show a sufficient range of diameters (Fig. 6) and moreover, appear adequately The colour, pH and organic content of soils from represented in the subcanopy layers (Fig. 4) to the six sites are given in Table 2. Colour showed justify designating this impressive forest stand a little variation either among the six sites or between climax community (Fig. 8). Trunks of rimn were the surface and a depth of 2 m. All samples are measured up to 90 cm diameter, kahikatea to 85 em, fairly uniformly acidic throughout the upper 2 m mira, which enters the sequence relatively late (Fig. although that beneath the mature forest is somewhat 4), to 80 em, kamahi to 75 em and silver beech to less so. The organic content exceeds 70 % in all but 70 cm. The relatively large size of the podocarp the lower samples of the three latest stages (sites trees is indicated by their greater percentage con- 4-6). In all samples the non-combustible material tribution to basal area than to density (Fig. 4). The is mostly coarse sand. Indeed, an almost pure, total density of trees (499/ha) is less than half that organically stained sand, presumably of alluvial recorded in either of the other two stands containing origin, was found underlying the mature forest at a trees, but the total basal area (6,557 dm'/ha) is depth of 1.3 m. Little profile development including similar to that recorded in the immature rimu stand gleying, is apparent at this or any other site, within (Stage 5) and substantially more than from the the upper 2 m. silver pine woodland (Stage 4). Even though density values for the woody sub- 3. Substrate Sampling of Peat Depth, Wood and canopy layers are less than those recorded from any Pol/en of the other stages, many more species are present The ground surface along a line at right angles to (see Appendix) and moreover, none are notably the pakihi margin extending to Site 1 (Fig. 3), slopes conspicuous. Minimal disturbance by deer is prob- very gently upwards (at ca. 0.23°) to an area about

TABLE2. Colour, pH and organic matter content (loss on ignition) of soils at two depths from the six sites described in the text and shown in Figure 3. Soil symbols and names follow the Munsell notation.

Site Depth pH %Org. Colour (em) matter No Vegetation

1 Pakihi 0-20 4.1 96.9 10YR 2/2 Very dark brown 180-200 4.2 95.6 10YR 2/2 " 2 Young Manuka 4.1 83.9 IOYR 2/2 0-20 " 10YR 2/2 180-200 4.2 90.7 " 3 Manuka Woodland 0-20 3.9 88.2 IOYR 2/2 " 180-200 4.1 76.1 IOYR 2/2 " Silver Pine Woodland IOYR 2/2 4 0-20 4.0 71.3 " 180-200 3.9 66.4 10YR 2/2 " 5 Young Rimu Stand 0-20 3.8 88.6 5YR 2/2 Dark red brown 180-200 4.1 59.5 5YR 2/2 " 6 Mixed Lowland Forest 0-20 4.4 72.0 10YR 2/2 Very dark brown 130-150 4.5 15.2 IOYR 3/2 Very dark gray brown 180-200 4.6 7.4 2.5YR 3/2 Very dark gray brown MARK AND SMITH: SOUTH WESTLAND LOWLAND VEGETATION; PART I 85

30 m out from the margin. Soundings of peat along ten probes intercepted wood and at a range of this line showed that its depth reaches a maximum depths. Of three specimens identified, one was (6.1 m) near the centre and becomes somewhat Dracophyllum and the other two were silver pine. thinner towards the margin (Fig. 9). Roots and This evidence for a previous woody cover on the site buried wood beneath the woody vegetation unfortun- is reinforced by the presence of occasional dead ately prevented the soundings being extended beyond charred stems of silver pine, 6-10 em diameter, still the bog margin. Where possible, samples of buried standing among the short open manuka. A similar wood were collected for identification but all occur- incidence of buried wood on the existing fringe rences have been shown on Figure 9. (Station 8 in Fig. 9) is also indicative of several All but one of the six wood samples encountered previous stands of woodland although only manuka in the peat beyond 30 ill from the pakihi fringe wood was identified from this station. were fragments and all were deeper than 2.4 m. Of Beneath the tall dense woodland of manuka the two fragments identified, one was of silver pine (Station 9 in Fig. 9) recognition of buried wood was while the other, a small twig, was juvenile wood of difficult because of the high density of roots but a kamahi (Fig. 9). It appears likely that all these six small specimen with bark still intact from 15 cm specimens would have been carried on to the bog, depth proved to be silver pine while two specimens perhaps during floods, but this explanation becomes of manuka wood were recovered 45 em down. Along less applicable as the incidence of buried wood the transition between manuka and silver pine wood- increases over the remaining 30 m to the existing lands there was a partly decayed surface log of margin of the bog. At Stations 5 and 6 (see Fig. 9), silver pine, one at 15 cm depth of bog pine (Dacry- 30 m and 20 m respectively from the margin, several dium bidwillii) and a large log, buried 40 em, which specimens were intercepted, but only at one or two proved to be yellow-silver pine (D. intermedium). levels, which again could be the result of periodic Wood from a depth of 45 em near the outer edge deposition by flood waters. At Station 7, however, of the silver pine zone was identified as "probably just 10 m out from the margin and within the zone Daerydium cupressinll1n" (R. N. PateJ, pers. cornm.). of young manuka (Stage 2 type), all but one of the Pollen in a sample of basal peat at 6.1 m depth

FIGURE 9. Profile from the silver pine woodland to the centre of the bog (Site 1 in Fig. 3) showing depth oj peat and incidence of buried wood at ten sites. Values for each site ha,,'e been spread out for clarity but are centred on the site numbers (circled). Only peat sample lines shown not ending in a wood specimen indicate the full depth of peat. Material for pollen analyses was taken from the site "P" at Station 1. Note, even though the bog surface is shown as level, it is actually highest at Site 5-from here it falls 30 cm to Site 8 and 55 em to Site 1. 86 PROCEEDINGS OF THE NEW ZEALAND EcOLOGICAL SOCIETY, VOL. 22, 1975

near the centre of the pakihi (Station 1 in Fig. 9), as community. well as that from the fine gray silt immediately undn- The vegetation sequence is accompanied by lying it, was anaJysed by Dr N. T. Maar. The results progressive changes both in canopy height and (Table II in Appendix), as interpreted by Dr Moar floristic richness, vascular species about trebling (peTs. comm.), suggest that the bottom peat is no in number between the bog (26) and the mixed older than 10,000 years B.P. Moreover, the site of forest (71). Subshrubs are most abundant in the deposition probably was not forested, since pollen silver pine woodland while shrubs reach their peak

of0- Phormium,.' Aj.telia, Cyperaceae and Gramineae in the young manuka along the fringe of the bog.

"suggest. ,", open,-damp-. conditions, with shrubs, e.g. Density of small-trees and trees, on the other hand, Leptospermum,. dotted . about. There is EttIe doubt, decreases steadily as an index of increasing size of however, that the region was forested at the time individuals beyond the woodland stages. There is by podocarps, and NQthofaRus menziesii, judged by an obvious steady increase in total biomass accom. the frequency of its pollen in the basal peat, was panying the sequence although no attempt was made probably locally abundant. PoIlen of the N. fusca to measure this. Even though the same values for type could have drifted in from a distance". total tree basal area were obtained for the young rimu forest and mixed forest stands the 10 m increase DISCUSSION AND CONCLUSIONS in canopy height would ensure a greater total biomass in the mixed forest. The,. .sequence of vegetation-pakihi bog to Since a pollen analysis from the deepest peat mixed beech-podocarp forest-surrounding Dismal sampled near the centre of the bog indicates a forest Swamp in South Westland. appears at least super- vegetation in the vicinity it seems unlikely that the fically to represent a genuine primary forest succes- bog was initiated at the close of the last Otiran sion. Thus the peat bog occupies a shallow, irregu- glacial advance about 14,000 years B.P. (Suggate and larly and imperfectly drained basin, perhaps an old Moar, 1970). Dr Moar (pers. comm.) assumes tnat main or flood channel of the Arawata River and, the basal peat is no oider than 10.000 years B.P. and except near its margins, contains no more buried suggests that the post-glacial transition from a wood than flood waters might have carried in. More- regional climax of shrubland to one of forest, well over, pollen analysis of its basal peat indicates that documented at Omoeroa Bluff, coast-wise of Franz the site probably was not forested although the Josef Glacier, may have been in the process of com- region apparently was (N. T. Moar, pers. comm.). pletion when the underlying mineral material In addition, patterns of size class distribution among sampled at Dismal Swamp was deposited. the woody species which achieve importance at some Although this vegetation sequence could represent stage along the vegetation sequence, consistently a genuine hydrarch succession the distinctive soil imply unidirectional and progressive changes towards profile underlying the mixed forest suggests that this the mixed beech-podocarp forest. Small shrubs of stand occupies a sandy rise on the edge of a silt- manuka and Dracophylluln longifolium emerge floored depression. In particular, the relatively shal- above the hummocky surface of the bog and towards low organic layer beneath the mixed forest indicates its margin, manuka in particular becomes more that it did not necessarily deveJop as the terminal numerous. taller and older. This open scrub merges phase of a succession represented by the other five with a close-canopied manuka thicket which increases sfands. There seems little doubt, however, that Stages in height to about 6 m with distance from the bog 1 to 5 (pakihi bog to young cimu forest) are related margin. Silver pine but not manuka seedlings are developmentally although fire has obviously delayed numerous beneath the dense manuka canopy and the early stages of this succession, Even without the two species may co-dominate a narrow transi- fire, however. the succession appears to be extremely tion to silver pine woodland containing overtopped slow, delayed undoubtedly by unfavourable soil con- and decadent stems of manuka. Rimu and kamahi, ditions, as suggested for pakihis generally by Hollo- already well established beneath the silver pine way (1954), and in particular to a high water table. canopy, eventually overtop it to become the most This is demonstrated both by the rapid establish- prominent members of the subclimax forest. Silver ment and growth of manuka along the embankment pine regeneration fails but many other forest species of the road traversing ODe arm of the bog and also establish. Eventually a mixed beech-podocarp forest by the presence in the manuka from the bog, of appears to develop. Quantitative data from this mixed inflated but highly porous bark at and below the forest indicate adequate replacement of all import~ soil surface. This tissue, which was also reported ant species and it is therefore interpreted as a climax recently from western Fiordland, may promote acra- MARK AND SMITH: SOUTH WESTLAND LOWLAND VEGETATrON; PART 1 tion of the roots (Wardle, et al., 1973). Growth rings in manuka. assum.ing they are annual, indicate that plants achieve rapid growth around the bog margin, relative to that of silver pine which eventually suc- ceeds it. Even within the tall manuka woodland, occasional co-dominant stems of silver pine may be five times as oM. This age anomaly appears to have resulted from the periodic destruction of woody vegetation along the margin of the bog. High incidence of buried wood of several woodland species and from a wide range of depths down to 3 m, within 20 m of the existing woodland margin, reveals that the bog margin has been subjected to several dis- turbances in the past. Even without disturbance, however, the very slow growth of silver pine suggests that several generations of manuka would FIGURE 10. The area shown in Figure 2 rephoto- occupy a site while silver pine builds up to domin- graphed about five years after the 1967-68 summer ance. But in the area studied, occasional charred, fire. Note destruction of all woody vegetation along standing small stems of silver pine among the open the forest margin and its replacement by an herbace- manuka in front of the dense woodland and close ous cover, chiefly Baumea teretifolia except for tall to the highway, indicates some importance of fire. tussock!,. of Gahnia rigida in the zone previously Information from local residents revealed that these occupied by woodland. October 1972. dead stems may have resulted from a fire during the 1930's which apparently was confined to a small may reach its southern limit on Dismal Swamp. area of manuka scrub close to the road. However, The relatively sharp boundaries between the silver the extent which fire can penetrate the woodland pine and manuka woodlands and between the closed was demonstrated during the 1967-68 summer when and open manuka stands along the margin of the most of the bog adjoining the highway on the east bog probably have resulted from previous fires whiCh was swept by a fire apparently lit intentionally to penetrated to varying extents. The other four pakihis facilitate the hunting of deer. This fire penetrated visited in the Dismal Swamp area were charactei'~ and destroyed aU of the manuka woodland surround- ised by both an increase in importance of woody ing Dismal Swamp and most of the silver pine wood- species and less abrupt margins than the one studied, land as well (Fig. 10). Many of the silver pine trees suggesting that fires have been less frequent in areas and all those of manuka perished, but by October more distant from the highway. Even so, the vegeta- 1972 manuka seedlings with up to five growth rings tion pattern is quite variable both within and between were plentiful throughout the destroyed woodland the pakihis shown in Figure 1, and a brief study so that another cycle appears to have been initiated. did not clarify the situation. The pakihi crossed by The fire also destroyed shrubs of manuka and Hindley Creek (see Figs. 1 and 3) obviously has Dracophyllum longifolium throughout the bog. In been free from fire for several decades at least. the initial post-fire period Baumea teretifolia made Shrubs of bog pine (Dacrydium bidwillii) ca I m the most rapid recovery but by 1972 both Calorophus tall are conspicuous among manuka and Dracophyl- and Gleichenia circinata had regained some of their /um near its margin (Fig. 11). Three stems of bog original importance. Most of the original species pine with basal diameters of 0.9 to 1.1 em contained were again present on the bog but still less important 39 to 42 growth rings. Bog pine persists even near than before the fire. Baumea teretifolia remained the the centre of this pakihi where Ca/orophus provides only vascular species to obviously increase in ca 70% of the cover and Gleichenia, Baunua spp. importance. and Lepidosperma australe are also common. Occa~ The general physiogomy of the pakihi is similar sional low shrubs of Dacrydium intermedium and to that described by Rigg (1962) near Westport since stunted stems of Libocedrus bidwi!lii up to 1.5 m the more important species are common to both tall, throughout much of the bog, also attest to the areas. However, 79 of the lOt vascular species, freedom from fire. including all eight exotics listed by Rigg. Were not Prevailing moisture conditions could determine the recorded at Dismal Swamp. One of the species extent of spread of a fire since vulnerability probably shared, the small bog eyebright Euphrasia disperma, decreases from bog via manuka to silver pine wood- 88 PROCEEDINGS OF THE NEW ZEALAND ECOLOGICAL SOCIETY, VOL. 22, 1975

Botany Division, DSIR, for analyses and interpreta- tion of pollen; Mr R. N. Patel" Forest Research Institute, N.Z.F.s., for assistance with wood identification; Mr G. A. H. Kidd, Geography Department, Uni- versity of Otago, for the preparation of the map; Mr J. R. Crush, Botany Department, University of Otago, for field assistance in 1972; the Surveying Department, University of Otago, for loan of surveying equipment; and the N'z. Meteorological Service for provision of climatological data.

REFERENCES ALLAN, H. H. 1961. Flora of New Zealand. Yol. I. Government Printer, Wellington, 1085 pp. FIGURE 11. View west towards the southwest corner CHAVASSE,C. G. R. 1962. Forests, soils and landforms of the pakihi crossed by Hindley Creek (see Figs. of Westland. New Zealand Forest Service Informa- I and 3). showing the vegetation pattern within ca tion Series No. 43. 100 In of its margin. In the foreground ~.hrubs ca CHEESEMAN, T. F. 1925. Manllal of the New Zealand J m tall of manuka, Dacrydium bidwillii and Draco- Flora. Government Printer, Wellington, 1163 pp. phyllum longjfolium with prominent tussocks of COCKAYNE,L. 1928. The VegetatIOn of New Zealand. Gahnia rigida (behind figure) extend above a herb 2nd ed. Engelmann, Leipzig, 456 pp. COTTAM, G.; CURTIS, J. T. 1956. The use of distance layer of Calorophus and Gleichenia circinata. With measures in phytosociological sampling. Ecology approach to the margin there are, in order, zones 37: 451-460. of manuka woodland. silver pine woodland and CRANWEI..L,L. M. 1952. An outline of New Zealand podocarp (mostly P. dacrydioides) forest. October peat deposits. Proceedings of the 7th Pacific Science 1972. Congress 5: 186-208. FRANKLIN,D. A. 1967. Basal area as determined by the land. Only exceptionaIIy, therefore, as with the recent point-centred quarter method. New Zealand Journal of Botany 5: 168-]69. fire. would fire penetrate to the Jatter. Protracted HOLLOWAY,J. T. 1954. Forests and climates in the fluctuations of the water table may also cause exten- South Island of New Zealand. Transactions of the sions and retractions of the woody vegetation along Royal Society of New Zealand 82: 329-410. the margin of a bog. The depth of wood remains McDoNAI"D. D. C. 1955. Soil moisture and physical pro- found suggests that at least some destruction occurred perties of a Westland "pakihi" soil in relation to prior to European settlement. deforestation. New Zealand Joumal of Science and A recommendation to the New Zealand Forest Technology B37: 258-266. Service and supported by the New Zealand Ecologi- MARK, A. F.; ESLER, A. E. 1970. An asses,sment of the point-centred quarter method of plotless sampling cal Society to reserve for educational and scientific in some New Zealand forests. Proceedings of the purposes, as well as for its aesthetic value, this New Zealand Ecological Society ]7: 106-110. sequence of South Wcestland lowland vegetation, has MOAR, N. T. 1955. Adventitious root-shoots of Dacry- been provisionally approved. The highway not only dium colensoi Hook in Westland, South Island, traverses the sequence but also separates the recently New Zealand. New Zealand JOt/rnal of Science and burnt eastern area from the unburnt areas to the west Technology A37: 207-213. of the highway. Local exploitation of silver pine from MOORE. L. B.; EDGAR, E. 1970. Flora of New Zealand. areas close to the highway on either side detracts Vol. II. Government Printer, Wellington, 354 pp. MUTCH, A. R.; McKELLAR, I. C. 1964. Sheet 19 Haast somewhat from its value as a reservation. However, (Ist ed.). Geological Map of New Zealand 1: extensive areas of uncut silver pine woodland persist 250,000. DSIR, Wellington, New Zealand. within the proposed reserve and moreover, the cut- RIGG, H. H. 1962. The pakihi bogs of Westport. New ting licence, which expired in 1972, apparently has Zealand. Transactions of the Royal Society of New not been renewed. Zealand (Botany) 1 (7): 91-108. SCUTT,G. A. M.; ROWLEY,JENNIFERA. 1975. A lowland ACKNOWLEDGEMENTS vegetation sequence in South Westland: Pakihi bog to mixed beech-podocarp forest Part 2. Ground and In addition to those who assisted with the main pro- epiphytic vegetation. Proceedings of New Zealand ject (see' Smith, 1965) we wish to thank Dr N. T. Moar, Ecological Socieiy 22: 93.108. MARK AND SMITH: SOUTH WESTLAND LOWLAND VEGETATION; PART 1 89

SMITH, P. M. F. 1965. Science Survey, Jacksons Bay, tation and landscape of the West Cape District, 1965: Botany; A preliminary report. Science Record Fiordland, New Zealand. New Zealand Journal of (Dunedin) 15: 63-67. Botany 11: 599-626. STEVENS, P. R. 1968. A chronosequence of soils near WELLS, 1. A'.; MARK, A. F. 1966. The altitudinal Franz Josef glacier. Unpub. Ph.D. thesis, Lincoln sequence of climax \'egetation on Mt Anglem, College, New Zealand. 'Stewart Island. Part I: The principal strata. New SUGGATE.R. P.; MOAR, N. T. 1970. Revision of the , Zealand Journal of Botany 4: 267-282. chronology of the late Otira Glacial. New Zealand ZOTOV, V. D. 1963. Synopsis of the grass subfamily Journal of Geology and Geophysics 13: 742-746. Arundinoideae in New Zealand. New Zealand W.\RDLE. P.; MARK, A. F.; BAYLIS, G. T. S. 1973. Vege~ Journal of Botany 1: 78-136. 90 PROCEEDINGS OF THE NEW ZEALAND ECOLOGICAL SOCIETY, VOL. 22, 1975

ApPENDIX TABLE 1. List of vascular plants present in one or more of the six stands representing the vegetation sequence from pakihi bog to low/and mixed beech-podocarp forest at Dismal Swamp, South West/and, Symbols denote the tallest layer occupied by the species in each stage as follows: T = tree; Sf ;= small-tree; Sh = shrub: Su = subshrub; H = herb: G = ground layer; E = epi- phyte; L = liane. An asterisk preceding the species name signifies ils occurrence in pakihis of fhe Westport area (Rigg, 1962). VEGETATION TYPE Bog Open Manuka Silver Young Mixed Manuka Woodland Pine Rimu Forest Woodland Forest Hemiphues suffocata var. novae zelandiae G - - - - - *Herpolirion novaezelandiae G - - - - - Lepidosperma australe H - - - - - *Liparophyllum gunnii G - - - - - Thelymitra venosa G G - - - - *Euphrasia disperma G - - - - - Prasophyllum colensoi G - - - - - *Drosera binata G - - - - - Dacrydium laxifolium G G - - - - Cyathodes empetrifolia H H - - - - *Schizaea fistulosa H H - - - - *Drosera spathulata G G - - - - *Centrolepis ciliata G G - - - - "'Utricularia monanthus G G - - - - Cassinia fulvida Sh Sh - - - - Coprosma brunnea Sh Sh - - - - "'Lycopodium ramulosum G G - - - - "'Notodanthonia nigricans H H H - - - "'Calorophus minor H H H H - - "'.Baumeateretifolia H H H H - - Baumea tenax H H H H - - Dracophyllum longifolium Sh Sh Sh Sh - - Dacrydium biforme - Su Sh St - - "'Gleichenia circinata H H H H H - "'Leptospermum scoparium Sh Sh T T Sh - "'Dacrydium colensoi - Su T T T - "'Dacrydium intermedium - - Sh T - - Libertia pulchella - - G G G - Gahnia rigida H H H H H - Nertera depressa G G G G G G "'.Blechnum minus - H H H H H Nertera dichondraefolia - G G G G G "'Dacrydium cupressinum - Su St T T T Neomyrtus pedunculata - Su Sh St St St Coprosma colensoi - Su Sh Sh Sh St Pseudopanax colensoi Philipson var. ternatus Wardle - Su St St T T Phyllocladus alpinus - Su Sh T T St Coprosma parviflora - Sh Sh Sh Sh Sh Coprosma foetidissima - Su Sh St St St Myrsine divaricata - Su Sh Sh St St "'Phormium tenax - - H - - - Gaultheria antipoda - - Sh - - - "'Metrosideros umbellata - - T T T - '" Blechnum capense - - H H H H Griselinia littoralis - - St St St T Weinmannia racemosa - - T T T T Cyathodes juniperina - - Sh Sh - - Podocarpus hallii - - Sh T T T Dacrycarpus dacrydioides (Rich.) deLaubenfel.\' - - Su Sh Sh T MARK AND SMITH: SOUTH WESTLAND LOWLAND VEGETATION; PART I 91

Coprosma ciliata - - Su Sh Su Su Coprosma lucida - - Su Sh Sh SI Clematis paniculata - - L L L L Earina autumnalis - - EH E E E Dendrobium cunoinghamii - - E E - E Hymenophyllum multifidum - - G E E E *Dianella intermedia - - H - - - Nothofagus solandri v. cliffortioides - - Sh T T SI Pseudopanax crassifoHus - - Sh SI SI T Grammitis heterophylla - - E E E E Hymenophyllum rarum - - E E E E Tmesipteris tannensis - - EG EG EG E Astelia nervosa - - H . H H H Carmichaelia grandiflora - - - SI - Pseudopanax simplex - - - Sh Sh Uncinia rupestris - - - H H H Gleichenia cunninghamii - - - H H H Luzuriaga parviflora - - - EG EG Libocedrus bidwillii - - - T - Nothofagus menziesii - - - T T T Elaeocarpus hookerianus - - - SI SI T Myrsine australis - - - Sh Sh SI Coprosma polymorpha - - - Sh Sh SI Grammitis billardieri - - - G G GE Asplenium ftaccidum - - - HE E E Dicksonia squarrosa - - - H H SI Trichomanes reniforme - - - EO EO EO Coprosma parviflora - - - Sh Sh Sh Podocarpus ferrugineus - - - SI Sh T Lindsaea trichomanoides - - - H H H Hymenophyllum revolutum - - - E E E Lycopodium volubile - - - - H Histiopteris incisa - - - - H Pittosporum crassicaule - - - - Sh Hedycarya arborea - - - - Su SI Microlaena avenacea - - - - H H Blechnum discolor - - - - H H Ascarina lucida - - - - Sh SI Lycopodium scariosum - - - - G 0 Lycopodium billardieri - - - - - E Pseudopanax edgerleyi - - - - - T Pseudopanax anomalus - - - - - Sh pseudowintera colorata - - - - - SI Carpodetus serratus - - - - - SI Astelia solandri - - - - - E Earina mucronata - - - - - E Scheffiera digitata - - - - - Sh Cyathea smithii - - - - - Sh Hymenophyllum scabrum - - - - - HE Hymenophyllum tlabellatum - - - - - E Rumohra adiantiformis - - - - - EH Hymenophyllum sanguinolentum - - - - - E Hymenophyllwn armstrongii - - - - - E Phymatodes diversifo1ium - - - - - HE Freycinetia banksii - - - - - L Ripogonum scandens - - - - - L Metrosideros diffusa - - - - - L Metrosideros fulgens - - - - - L Metrosideros perforata - - - - - L Rubus australis - - - - - L Me1icytus ramitlorus - - - - - SI Coprosma rotundifo1ia - - - - - Sh Gahnia procera - - - - - H TOTAL NUMBER 26 31 45 59 57 n 92 PROCEEDINGS OF THE NEW ZEALAND ECOLOGICAL SOCIETY, VOL. 22, 1975

TABLEII. Pol/en analyses of two samples from a pakihi bog, Dismal Swamp, South Westland. Values are percentage figures based on total tree and shrub pollen counted. Analyses by Dr. N. T. Moar.

Underlying Basal peat mineral c. 6.1 m depth material

Dacrycarpus dacrydioides 7 5 Podocarplls spp. (incl. P. jerrugineus, P. spicatlls and P. totara) 18 17 Dacrydium cupressinum 31 30 D. hidwillii-hijorme I 2 Phyl/oc/adus 2 3 Nothofagus fusca type 3 N. menziesii 9 Ascarino 3 2 Coprosma 8 19 Coriaria 2 - Hoheria I 2 Leguminosae - 2 Leptospermum scoparium I 3 Metrosideros 2 - Myrtaceae I 2 Myrsine 7 3 Pseudopanax 3 - Astelia I 3 Chenopodiaceae Ir. Compositae I Cyperaceae 33 24 Gramineae 5 19 Gunnera Ir. 16 Halorogis - 2 Hydrocoty/e Ir. Rumex I Rosaceae Ir. Phormium 6 5 Mondcte fern spores 29 92 Phyma/odes dil'ersifolium 4 6 Trilete fern spores 2 2 Cyathea colensoi type 41 40 DicksoniQ squarrosa 2 - G/eichenia circinata 15 2 Lycopodium 10 3 Sphagnum 30 50 Total tree and shrub pollen 262 62